BUILDING RESEARCH JOURNAL
DOI:10.2478/brj-2014-0001
VOLUME 61 (2014), NUMBER 1, pp. 1 – 14
PREDICTING THE DURATION OF CONCRETE OPERATIONS VIA
ARTIFICIAL NEURAL NETWORK AND BY FOCUSING ON SUPPLY
CHAIN PARAMETERS
MOJTABA MAGHREBI 1, CLAUDE SAMMUT 2 and TRAVIS S. WALLER 1
Being able to precisely predict the duration of concrete operations can help construction managers
to organize sites and machineries more efficiently, especially when there is limited space for
equipment on site. Currently there is no theoretical method for estimating the duration of the
concrete pouring process. Normally, the maximum capacity of pumping facilities on construction
sites is not used, and concrete pumps are idle for a considerable time as a result of the arrival of
concrete trucks being delayed. In the light of this issue, this paper considers the supply chain
parameters of Ready Mixed Concrete (RMC) as a means of solving this problem. Artificial Neural
Network (ANN) is hired for modelling/predicting the productivity of a concrete operation. The
proposed model is tested with a real database of an RMC in the Sydney metropolitan area that has
17 depots and around 200 trucks. Results show that there is an improvement in the achieved results
when these are compared to the results of relevant studies that only considered the construction
parameters for predicting the productivity of concrete operations.
Keywords: RMC, productivity, supply chain
1. Introduction
In construction, concrete plays a key role as the world`s most used construction material
(Kosmatka et al., 2002). It is also expected that the demand for concrete will increase in the
future (Imbabi et al., 2012, Council, 2009, Humphreys and Mahasenan, 2002, Damtoft et al.,
2008, Rosenthal, 2007, Mahasenan et al., 2003, Worrell et al., 2001, Mehta, 2009) This issue is
studied in detail in the report published by International Cement Review (ICR) (Armstrong
2013). They reviewed the cement market in 165 countries over 22 years. This comprehensive
study emphasizes that regardless to the geographical location the demand for concrete increase
globally. Therefore, it is a necessity that the ready mixed concrete (RMC) industry is enhanced
to cope with higher demands.
Most of the publications in the RMC domain have been devoted to implementing heuristic
methods which Genetic Algorithm (GA) has been highlighted more than other heuristic
methods. Garcia et al. (2002) modelled the RMC for a single depot and solved it via
optimization and GA. However, their approach is not practical because some realistic
constraints were relaxed and considered only small instances. (Feng et al., 2004) also modelled
a single depot RMC and assumed some parameters such as loading/unloading times as fixed
1
The University of New South Wales (UNSW), School of Civil and Environmental Engineering, Sydney, Australia,
e-mail: [email protected]
2
The University of New South Wales (UNSW), School of Computer Science Engineering Sydney, Australia
Unauthenticated
Download Date | 7/31/17 11:41 PM
1
Mojtaba Maghrebi, Claude Sammut and Travis S. Waller
parameters. The instances that have been considered by them are much smaller than the
instances that are used in this paper. Naso et al. (2007) modelled a more realistic RMC problem
by considering multi-depots and penalizing the waiting times (loading/unloading) in the
objective function. They also introduced a GA algorithm which is very similar to the methods
that were presented earlier by Garcia et al. (2002) and Feng et al. (2004). However, the
instances that Naso and colleagues have tested are larger than in the previous research. Lu (2002
and 2003) developed a software package called HCKCONSIM to deal with real RMC problems.
This mainly concerned the discrete event simulation (DES) tool but in its recent versions was
coupled with heuristic solvers such as GA (Cao et al., 2004, Lu and Lam, 2005, Ming and HoiChing, 2009), Particle Swarm Optimization (PSO) (Lu et al., 2006, Wu et al., 2005) and real
GPS data of trucks (Lu et al., 2007) in order to make a more powerful tool. Feng and Wu (2006)
and Cheng and Yan (2009) had a similar approach by integrating DES with a fast messy GA
algorithm. Silva et al. (2005) compared GA with Ant Colony Optimization (ACO) and
suggested a GA-ACO method for solving RMC problems. Pan et al. (2010) proposed an
improved Discrete PSO (DPSO) for solving RMC dispatching problems and recently
Srichandum and Rujirayanyong (2010) compared Bee Colony Optimization (BCO) and Tabu
Search (TS) with GA in this context. Despite developments in this area, the solution structure
among most introduced methods is pretty much same, especially in the GA based method where
the chromosome structure consists of two merged parts: the first part defines the sources of
deliveries; the second part expresses the priorities of customers. The solution structure in these
techniques is quite simple and easy to understand. However, a cumbersome computing process
must be completed in each iteration to check the constraints or after achieving a premature
solution.
In the literature, rather than GA some other approaches also have been studied that will be
discussed briefly in the text that follows. Yan et al. (2008) introduced a numerical method for
solving the RMC optimization problem by cutting the solution space and incorporating the
branch and bound technique and the linear programming method. Lin et al. (2010) modelled the
RMC as a job shop problem. Yan et al. (2012) used decomposition and relaxation techniques
coupled with a mathematical solver to solve the problem, and Payr and Schmid (2009) applied
Variable Neighborhood Search (VNS) to deal with RMC optimization problems. Asbach et al.
(2009) made the mathematical modelling much simpler by dividing the depots and customers
into sub-depots and sub-customers. They also used large scale instances for testing their
introduced large neighborhood search and decomposition methods.
It has been found that of the related works in the literature about RMC only Graham et al
(Graham et al., 2006) consider this issue in detail. The concern of these authors is in predicting
the duration of concrete operations accurately. In this regard they used Artificial Neural
Network (ANN) and tested it via a real database. They assumed that the duration of an operation
is related to the construction situation. Therefore, they collected data from four actual
construction projects which consisted of the following attributes: month of operation, type of
operation, truck volume, total operation volume, average interarrival time, number of loads in
operation, number of accepted loads and number of rejected loads. It appears that they assumed
that there is no limitation for RMCs and the required trucks usually arrive at construction sites
without any delay, which in reality is not completely true. Graham et al (Graham et al., 2006)
indirectly assumed that the duration is related to the type of construction operation (wall,
column, slab), truck volume, interarrival time and ... etc.
However, in the authors’ opinion the RMC conditions have a significant impact on the
duration of the operation. In other words, environmental variables must also be taken into
2
Unauthenticated
Download Date | 7/31/17 11:41 PM
Predicting the duration of concrete operations via artificial neural network
and by focusing on supply chain parameters
account. For example, the duration of a concrete pouring task in a project which is located in a
city with dense traffic would be different from one in a small city with far less traffic.
Furthermore, the travel time between a depot and a project in the early morning varies from the
travel time at midday. Travel times from/to some locations during the day vary considerably
although they are less critical on non-working days. The importance of a quality prediction
emerges when a concrete pump and workers are around 12 % idle on the site, making for an
additional 14 % cost (Graham et al., 2006). Consequently, it is necessary to have a more
accurate prediction of the duration of the operation.
The second contribution of this paper is the size and amount of data that is used in this paper
which is much greater than the datasets that have been used in similar research in literature. The
richness of data helps the authors to conclude confidently.
2. Methodology
Typically, fresh concrete is hauled by trucks from batch plants to the construction sites and
then placed in frames to construct concrete elements. A project might need several deliveries;
therefore the required trucks must arrive at the site consecutively. However, there is no general
model in the literature that can predict the duration of a concrete operation precisely.
It is recommended to take into consideration the both internal and external parameters. It is
not intended to check the available data with several algorithms in order to find which algorithm
outperforms others. Although this would be valuable and the researchers might consider doing
this in the future, in this paper the authors want to place more emphasis on how machine
learning techniques – and specifically ANN – can be hired in construction management more
effectively. In this paper, then, the problem is considered in general terms and from the
perspective of both clients and RMCs. In this regard, first the proposed attributes are justified.
Second, the ANN model is introduced. Third, the features of the available database are
examined. Finally, the achieved results are discussed and compared with relevant research. It is
worthy to note that authors do not intend to estimate the size of crew or machinery for concrete
pouring which has been discussed in literature extensively (Thomas and Sakarcan, 1994,
Thomas et al., 1984, Thomas and Daily, 1983, Thomas, 1991, Sonmez and Rowings, 1998,
Crawford and Vogl, 2006, Borcherding and Alarcon, 1991, Zahraee et al., 2013, Dunlop and
Smith, 2004). Beyond the construction sites there are important variables that have not been
carefully taken into account for predicting the productivity of concrete pouring. These variables
will be associated into the proposed ANN based model to predict the duration of concrete
pouring effectively.
3. Data structure
As mentioned above, this paper aims to consider both the execution and supply chain
parameters related to concrete operation in the modelling process. It is recommended that a
database contains the following parameters which cover the affective parameters of the duration
of the operation.
– Weekday: The travel time for some areas is considerably different on working days and nonworking days. A digit between 1 and 7 is assigned for each day of week. For example,
Monday = 1 and Friday = 5.
– Starting Time of First Delivery: The duration of an operation would vary depending, for
instance, on whether it commences during a rush hour or at midnight. In this regard, the time
of arrival of the first truck to the site is extracted from the database. The ANN accepts
Unauthenticated
Download Date | 7/31/17 11:41 PM
3
Mojtaba Maghrebi, Claude Sammut and Travis S. Waller
nominal attribute so the time must be converted to the real number, for example 13.75 instead
of 1:45 pm.
– Total Amount of Ordered Concrete: For each project the total amount of delivered concrete
is extracted from the available data. This is expressed as a real number with one decimal.
– Location of Project: Predicting travel times cannot rely too much on the distance alone
because the speeds of trucks on some routes fluctuate in the course of a day. Therefore, an
expected arrival time based only on distance cannot be precise. Moreover, some parts of
metropolitan areas have different traffic patterns during day. The authors believe that geolocation data that includes longitude and latitude can possibly convey this information. Each
location (depots or projects) has unique longitude and latitude that are extracted from
available database with arithmetic precision to six digits. Thus, it is expected that these
provide enough information for the algorithms to determine the reasonable correlation
between geo-locations and other attributes. Possibly both attributes of the “Starting Time of
First Delivery” and “Location of Project” can in conjunction deliver the traffic pattern for
each location which would have significant impact on travel times.
– Total Number of Received Orders by RMC: This attribute becomes important when an
RMC has accepted numerous deliveries in a day. Possibly for some hours the available
resources of the RMC are not sufficient and demand is greater than supply. In such situations
RMCs stretch the interarrival times to balance demand and supply. This makes supplying
some deliveries possible although perhaps later than expected. Thus, this attribute can reflect
how busy the RMC is.
– Total Number of Assigned Deliveries to the Source Depot: The former attribute shows the
density of orders through the day; however, this attribute can reflect the same issue but
particularly for the allocated depot which is chosen to supply concrete to the project. This
attribute is selected when, for instance, an RMC has received many orders but demands are
not distributed among the supply area smoothly. In the other words, a depot can have a large
number of orders in some areas but very few orders in other areas; in such cases it is expected
that this attribute will assist ANN to realize this issue. For large projects with more than 30
deliveries, normally concrete is supplied from more than one depot. However, based on the
supplied records we still can recognize a depot as the main depot in database.
– Productivity: This is calculated by dividing the total amount per duration; its unit is m3/hr.
As it was mentioned the proposed method will not associate all the construction site
parameters directly in the model. For example: the proposed model does not associate the
pouring system (crane or pump) or type of operation (wall, column or base) in the calculation.
4. Artificial neural network
The concept underlying Artificial Neural Network (ANN) was inspired by the biological
nervous system (McCulloch and Pitts, 1943). A considerable amount of literature has been
published on ANN and its applications; however, in this paper we will discuss only some ANN
applications in construction management.
Moselhi and colleagues are among the first scholars to research ANN as a promising
management tool in construction (Moselhi et al., 1991). Following on from their work, Savin et
al. (1996) used ANN for resource leveling. The implementation of ANN for pavement
management was conducted by Brega et al. (1998). A large and growing body of literature has
investigated the prediction of earthmoving process by ANN such as (Shi, 1999, Chao and
4
Unauthenticated
Download Date | 7/31/17 11:41 PM
Predicting the duration of concrete operations via artificial neural network
and by focusing on supply chain parameters
Skibniewski, 1994, Kartam, 1996, Flood and Christophilos, 1996) and similarly for concrete
framework production (Portas and AbouRizk, 1997). ANN has also been implemented in
predicting complex issues in construction such as labour productivity (AbouRizk and Wales,
1997, Lu et al., 2000). Cost estimations on reconstruction projects via ANN was discussed by
Attalla and Hegazy (2003). Enhancing ANN with fuzzy logic and its application in estimating
contractors` markup has been studied by Liu and Ling (2003). Chehayeb et al. (2007) studied
the classification of construction court cases in Canada by ANN. Ko and Cheng (2007) fused
genetic algorithms, fuzzy logic and ANN to predict the success of construction projects
dynamically. Chen and Hsu (2007) introduced a method for solving potential lawsuit problems
as a result of changing orders via merging ANN and case-based reasoning (CBR). Shi and Li
(2008) adopted particle swarm optimization (PSO) in ANN for assessing and predicting the
quality of construction projects in China. Predicting the performance of international
construction projects was studied by Kim et al. (2009). Cheng et al. (2009) proposed an ANNbased model for achieving strategic control over project cash flows. Jin and Zhang (2011)
suggested an ANN-based model to model the risk allocation in public private partnership (PPP)
projects. The application of ANN in knowledge management in construction firms was studied
by Kale and Karaman (2011). Kim (2013) offered technique for accurately estimating the
highway project in South Korea by analytic hierarchy process (AHP), CBR and ANN.
It can be deduced from the applications of ANN in construction that this algorithm is a capable
tool for predicting complex processes and tasks. Other scholars have claimed that ANN is an
adaptive learning algorithm (Darrat and Zhong, 2000, Szu et al., 1992, Fast and Palme, 2010,
Qin et al., 2013, Zor et al., 2012), which means it is capable of finding a relationship between
inputs and outputs.
In this paper, MATLAB neural network fitting tool is used for training, validation and testing
the proposed problem. MATLAB uses the Levenberg-Marquardt (LM) algorithm (Levenberg,
1944, Marquardt, 1963) which is a backpropagation ANN. LM is originally a numerical method
for finding the minimum of a function, especially the least square curve fitting problems. All the
computational related processes have been developed in MATLAB with a core duo 3.00 GHz
processor, 8.00 GB RAM on Windows 7 Enterprise.
As mentioned above, the aim is not to find the best algorithm for predicting the productivity
among the machine learning algorithms. Nevertheless, according to the achievements of
Graham et al (Graham et al., 2006), LM is a proper choice for modelling and predicting the
productivity of the concrete pouring process. This algorithm is used in this paper for the same
objective but with a totally different data structure than has been proposed in the prior section.
The architecture of ANN is illustrated in Figure 1. This kind of feed-forward ANN algorithm
has connections between side layers. In this paper the number of hidden layers is selected from
2 to 10. A slight improvement in results was observed when the number of hidden layers was
increased; however, the differences are not significant. Therefore, in this paper only the results
of ANN model with 10 hidden layers are reported. MSE and R2 (coefficient of determination)
are the most important parameters which are used for measuring the quality of predictions.
MSE
(1)
When:
n = the number of instances
= Target (actual) value of instances i
= Predicted value for instances i
Unauthenticated
Download Date | 7/31/17 11:41 PM
5
Mojtaba Maghrebi, Claude Sammut and Travis S. Waller
The coefficient of determination (R2) is another assessing parameter that measures the
correlation between targets and predicts among the database. Its calculation is mentioned in
(eq. 2).
1
∑
∑
(2)
When:
T
∑
(3)
Figure 1. Architecture of ANN model
5. Data collection
For testing the proposed ANN model a real database which covers all deliveries of an RMC
for a period of 4 months was used. The dataset belongs to the Sydney metropolitan area, which
is the biggest city in Australia. It is a useful database for testing the proposed ANN model
because it covers many different areas with different traffic patterns. To understand the size of
an RMC we can briefly say that on 109 days the RMC was active and supplied 42793 deliveries.
The RMC had 17 batch plants and 217 trucks. On about 80 % of days the RMC delivered more
than 300 trucks per day. This number on around 50 % of days is more than 500.
6
Unauthenticated
Download Date | 7/31/17 11:41 PM
Predicting the duration of concrete operations via artificial neural network
and by focusing on supply chain parameters
Most projects need fewer than 5 deliveries and the number of projects that need only one
delivery is considerably large. Predicting the productivity is not very important for this size of
project; therefore, it focuses only on projects that order more than 5 deliveries on a day.
Consequently, 1673 projects were selected for further studies. Studies of RMC problems with
this size of database have not been conducted in the literature and it helps the researcher to be
more confident about covering the variety of possible situations and scenarios among the
collected data. The suggested attributes were obtained from the chosen projects and the
training/test set was constructed. In the following section the achieved results are discussed.
6. Results
The number of hidden layers as mentioned above is set to 10; 70 % of data (1171 instances)
is used for training and 15 % (251 instances) for validation, with the remaining 15 % used for
testing the ANN model. MATLAB R2012a automatically scales input values to [0 1] so it does
not need to do this job manually. Through the validation the ANN parameters are tuned to find
the optimum model by monitoring MSE (Mean Squared Error) (eq. 1) after any changes.
This validation process is very important because it can avoid over fit in the model. This can
happen when the ANN model is fitted to the training dataset precisely, thereby losing
generalization (Nissen and Nemerson, 2000). For this reason the model is tested with unseen
data through validation to make judgment about the trained model. Epoch is one of the critical
parameters that are adjusted in validation. An epoch is one training step when all instances of
the training dataset are shown in the ANN model. The result of adjusting an epoch is shown in
Figure 2, and as it is obvious the best epoch is 5 when the minimum MSE of 10.94 is achieved
for validation dataset.
Figure 2. Performance of the ANN model in validation process
The achieved results of training, validation and testing are depicted in Figure 3. The desired
behaviour occurs when ANN predictions are exactly equal to actual values. The obtained MSE
for training, validation and testing subsequently are 10.34, 10.94 and 8.83. The regression R
values also subsequently are 0.79, 0.77 and 0.81. The summary of all 1673 instances is shown in
Figure 4. From this graph we cannot deduce a trend for errors of the ANN model, therefore the
histogram of errors is drawn in Figure 5. This graph shows that, despite the number of
disproportionate errors being not too great, the errors larger than 10 belong to the training set.
Unauthenticated
Download Date | 7/31/17 11:41 PM
7
Mojtaba Maghrebi, Claude Sammut and Travis S. Waller
Another important issue that we can comprehend from Figure 5 is that around 70 % of training
errors and 80 % of validation errors and test errors are in the range of – 2.85, 2.18. This reflects
that the ANN model is able to predict good enough values in most of instances. In comparison,
the achieved results to the only similar research that was conducted by Graham et al. (2006)
show an improvement is obtained. In the best case they achieve 10.76 for MSE while in this
paper the 8.83 is achieved for MSE of test dataset and MSE in less than 10 in overall. The only
concern here is in relation to the possible correlation between errors and predictions. The
residuals for all instances are demonstrated in Figure 6. According to the graph there is no any
obvious trend or correlation between errors and targets. Most of productivities are in range of
8 – 15 (m3/hr) which the ANN model predicts those precisely, however for productivities less
than 5 (m3/hr) and more than 15 (m3/hr) the distribution of residuals are expanded gradually.
Although that there is no clear trend through this randomly growth, however, this issue can
possibly be studied in the future to understand the underlying reasons for this behavior.
Generally it is restated that the proposed ANN model is capable to predict productivity of
concrete operation with a better accuracy than relevant studies by considering both construction
and supply chain parameters.
Figure 3. Achieved results of the ANN model in training, validation and test
Figure 4. Summary of all instances
8
Unauthenticated
Download Date | 7/31/17 11:41 PM
Predicting the duration of concrete operations via artificial neural network
and by focusing on supply chain parameters
Figure 5. Errors histogram
Figure 6. Residual of all instances
Unauthenticated
Download Date | 7/31/17 11:41 PM
9
Mojtaba Maghrebi, Claude Sammut and Travis S. Waller
Predicting the productivity of pouring concrete in large projects is a challenging issue
because there are a large number of effective variables and also an absence of any theoretical
method. Graham et al. (2006) solved this problem via Artificial Neural Network (ANN) by
building a database focusing more on construction attributes. In this paper, ANN was hired as
well but totally different attributes were applied. In the authors’ opinion in mega cities, such as
Sydney that has been selected for case study in this paper, the supply chain issues are more
critical than the construction parameters. Therefore, by considering these issues a set of
attributes can be proposed based on the required data that were collected. The available database
consists of 4 months’ deliveries of a RMC with 17 depots and more than 200 trucks in the
Sydney metropolitan area. The size of the collected database was greater than in similar
research in order to cover all real possibilities and scenarios. Although that the size of data
which was used in this paper is larger than datasets that have been used in similar approach,
however, the achieved results indicate that the proposed ANN model works slightly better than
very similar model introduced by Graham et al. (2006) model. This model can be used by both
RMCs and clients for predicting the duration of concrete operations and thereby reducing the
idleness and cost of equipment in construction sites.
REFERENCES
[1] Abourizk, S. M. and Wales, R. J. (1997), “Combined discrete-event/continuous simulation
for project planning”, Journal of construction engineering and management 123, 11-20.
[2] Armstrong , T. (2013), “The Global Cement Report”, In: 10TH (ed.), International Cement
Review.
[3] Asbach, L., Dorndorf, U. and Pesch, E. (2009), “Analysis, modeling and solution of the
concrete delivery problem”, European Journal of Operational Research 193, 820-835.
[4] Attalla, M. and Hegazy, T. (2003), “Predicting cost deviation in reconstruction projects:
Artificial neural network versus regression”, Journal of Construction Engineering and
Management-Asce 129, 405-411.
[5]
Borcherding, J. D. and Alarcon, L. F. (1991), “Quantitative effects on construction
productivity”, Constr. Law. 11, 1.
[6] Brega, J. R. F., Soria, M. H. A., Marar, J. F. and Sementille, A. C. (1998), An intelligent
system for pavement management. In: Rogers, S. K., Fogel, D. B., Bezdek, J. C. and
Bosacchi, B. (eds.) Applications and Science of Computational Intelligence. Bellingham:
Spie-Int Soc Optical Engineering.
[7] Cao, M., Lu, M. and Zhang, J.-P. (2004), “Concrete plant operations optimization using
combined simulation and genetic algorithms”, Machine Learning and Cybernetics,
Proceedings of 2004 International Conference on, 2004. IEEE, 4204-4209.
[8] Chao, L.-C. and Skibniewski, M. J. (1994), “Estimating construction productivity: Neuralnetwork-based approach”, Journal of Computing in Civil Engineering 8, 234-251.
10
Unauthenticated
Download Date | 7/31/17 11:41 PM
Predicting the duration of concrete operations via artificial neural network
and by focusing on supply chain parameters
[9] Chehayeb, A., Al-Hussein, M. and Flynn, P. (2007), “An integrated methodology for
collecting, classifying, and analyzing Canadian construction court cases”, Canadian Journal
of Civil Engineering 34, 177-188.
[10] Chen, J. H. and Hsu, S. C. (2007), “Hybrid ANN-CBR model for disputed change orders in
construction projects”, Automation in Construction 17, 56-64.
[11] Cheng, M. Y., Tsai, H. C. and Liu, C. L. (2009) “Artificial intelligence approaches to
achieve strategic control over project cash flows”, Automation in Construction 18, 386393.
[12] Cheng, T. M. and Yan, R. Z. (2009), “Integrating messy genetic algorithms and simulation
to optimize resource utilization”, Computer‐Aided Civil and Infrastructure Engineering 24,
401-415.
[13] Council, W. B. (2009), World Business Council for Sustainable Development/The Cement
Sustainability Initiative, Cement Industry Energy and CO2 Performance: Getting the
Numbers Right.
[14] Crawford, P. and VOGL, B. (2006), “Measuring productivity in the construction industry”,
Building Research & Information 34, 208-219.
[15] Damtoft, J., Lukasik, J., Herfort, D., Sorrentino, D. and Gartner, E. (2008), “Sustainable
development and climate change initiatives”, Cement and concrete research 38, 115-127.
[16] Darrat, A. F. and Zhong, M. (2000) “On Testing the Random‐Walk Hypothesis: A Model‐
Comparison Approach”, Financial Review 35, 105-124.
[17] Dunlop, P. and Smith, S. D. (2004), “Planning, estimation and productivity in the lean
concrete pour”, Engineering, Construction and Architectural Management 11, 55-64.
[18] Fast, M. and Palme, T. (2010), “Application of artificial neural networks to the condition
monitoring and diagnosis of a combined heat and power plant”, Energy 35, 1114-1120.
[19] Feng, C.-W., Cheng, T.-M. and Wu, H.-T. (2004), “Optimizing the schedule of dispatching
RMC trucks through genetic algorithms”, Automation in Construction 13, 327-340.
[20] Feng, C.-W. and Wu, H.-T. (2006), “Integrating fmGA and CYCLONE to optimize the
schedule of dispatching RMC trucks”, Automation in Construction 15, 186-199.
[21] Flood, I. and Christophilos, P. (1996), “Modeling construction processes using artificial
neural networks”, Automation in construction 4, 307-320.
[22] Garcia, J., Lozano, S., Smith, K., Kwok, T. and Villa, G. (2002), “Coordinated scheduling
of production and delivery from multiple plants and with time windows using genetic
algorithms, Neural Information Processing”, ICONIP'02, Proceedings of the 9th
International Conference on, IEEE, 1153-1158.
[23] Graham, L. D., Forbes, D. R. and Smith, S. D. (2006), “Modeling the ready mixed concrete
delivery system with neural networks”, Automation in construction 15, 656-663.
[24] Humphreys, K. and Mahasenan, M. (2002) Towards a sustainable cement industry, Climate
change, sub-study 8, World Business Council for Sustainable Development.
Unauthenticated
Download Date | 7/31/17 11:41 PM
11
Mojtaba Maghrebi, Claude Sammut and Travis S. Waller
[25] Imbabi, M. S., Carrigan, C. and Mckenna, S. (2012), “Trends and developments in green
cement and concrete technology”, International Journal of Sustainable Built Environment
1, 194-216.
[26] Jin, X. H. and Zhang, G. M. (2011), ”Modelling optimal risk allocation in PPP projects
using artificial neural networks”, International Journal of Project Management 29, 591-603.
[27] Kale, S. and Karaman, E. A. (2011), “Evaluating the Knowledge Management Practices of
Construction Firms by Using Importance-Comparative Performance Analysis Maps”,
Journal of Construction Engineering and Management-Asce 137, 1142-1152.
[28] Kartam, N. (1996), “Neural Network–Spreadsheet Integration for Earth‐Moving
Operations”, Computer‐Aided Civil and Infrastructure Engineering 11, 283-288.
[29] Kim, D. Y., Han, S. H., Kim, H. and Park, H. (2009), “Structuring the prediction model of
project performance for international construction projects: A comparative analysis”,
Expert Systems with Applications 36, 1961-1971.
[30] Kim, S. (2013), “Hybrid forecasting system based on case-based reasoning and analytic
hierarchy process for cost estimation”, Journal of Civil Engineering and Management 19,
86-96.
[31] Ko, C. H. and Cheng, M. Y. (2007), “Dynamic prediction of project success using artificial
intelligence”, Journal of Construction Engineering and Management-Asce 133, 316-324.
[32] Kosmatka, S. H., Panarese, W. C., Allen, G. E. and Cumming, S. (2002), Design and
control of concrete mixtures, Portland Cement Association Skokie, Ill.
[33] Levenberg, K. (1944) “A method for the solution of certain problems in least squares”,
Quarterly of applied mathematics 2, 164-168.
[34] Lin, P.-C., Wang, J., Huang, S.-H. and Wang, Y.-T. (2010), “Dispatching ready mixed
concrete trucks under demand postponement and weight limit regulation”, Automation in
Construction 19, 798-807.
[35] Liu, M. and Ling, Y. Y. (2003), “Using fuzzy neural network approach to estimate
contractors' markup”, Building and Environment 38, 1303-1308.
[36] Lu, M., Abourizk, S. and Hermann, U. H. (2000), “Estimating labor productivity using
probability inference neural network”, Journal of Computing in Civil Engineering 14, 241248.
[37] Lu, M., Dai, F. and Chen, W. (2007), “Real-time decision support for planning concrete
plant operations enabled by integrating vehicle tracking technology, simulation, and
optimization algorithms”, Canadian Journal of Civil Engineering 34, 912-922.
[38] Lu, M. and Lam, H.-C. (2005), “Optimized concrete delivery scheduling using combined
simulation and genetic algorithms”, Proceedings of the 37th conference on Winter
simulation, Winter Simulation Conference, 2572-2580.
[39] Lu, M., Wu, D.-P. and Zhang, J.-P. (2006), A particle swarm optimization-based approach
to tackling simulation optimization of stochastic, large-scale and complex systems,
Advances in Machine Learning and Cybernetics, Springer.
12
Unauthenticated
Download Date | 7/31/17 11:41 PM
Predicting the duration of concrete operations via artificial neural network
and by focusing on supply chain parameters
[40] Mahasenan, N., Smith, S., Humphreys, K. and Kaya, Y. (2003), “The cement industry and
global climate change: current and potential future cement industry CO2 emissions”,
Greenhouse Gas Control Technologies-6th International Conference, Oxford: Pergamon,
995-1000.
[41] Marquardt, D. W. (1963), “An algorithm for least-squares estimation of nonlinear
parameters”, Journal of the Society for Industrial & Applied Mathematics 11, 431-441.
[42] McCulloch, W. S. and Pitts, W. (1943), “A logical calculus of the ideas immanent in
nervous activity”, Bulletin of mathematical biology 5, 115-133.
[43] Mehta, P. K. (2009), “Global concrete industry sustainability”, Concrete international 31,
4.
[44] Ming, L. and Hoi-Ching, L. (2009), “Simulation-optimization integrated approach to
planning ready mixed concrete production and delivery: Validation and applications”,
Simulation Conference (WSC), Proceedings of the 2009 Winter, 13-16 Dec. 2009. 25932604.
[45] Moselhi, O., Hegazy, T. and Fazio, P. (1991), “Neural networks as tools in construction”,
Journal of Construction Engineering and Management-Asce 117, 606-625.
[46] Naso, D., Surico, M., Turchiano, B. and Kaymak, U. (2007), “Genetic algorithms for
supply-chain scheduling: A case study in the distribution of ready-mixed concrete”,
European Journal of Operational Research 177, 2069-2099.
[47] Nissen, S. and Nemerson, E. (2000), Fast artificial neural network library, Available at
leenissen. dk/fann/html/files/fann-h. html.
[48] Pan, L., Liya, W., Xihai, D. and Xiang, G. (2010), “Scheduling of dispatching Ready
Mixed Concrete trucks trough Discrete Particle Swarm Optimization”, Systems Man and
Cybernetics (SMC), IEEE International Conference on, 10-13 Oct. 2010. 4086-4090.
[49] Payr, F. and Schmid, V. (2009), “Optimizing Deliveries of Ready-Mixed Concrete”,
Logistics and Industrial Informatics, LINDI 2009, 2nd International, IEEE, 1-6.
[50] Portas, J. and Abourizk, S. (1997), “Neural network model for estimating construction
productivity”, Journal of construction engineering and management 123, 399-410.
[51] Qin, Q., Wang, Q.-G., Li, J. and Ge, S. S. (2013), “Linear and Nonlinear Trading Models
with Gradient Boosted Random Forests and Application to Singapore Stock Market”,
Journal of Intelligent Learning Systems and Applications 5, 1-10.
[52] Rosenthal, E. (2007), “Cement industry is at center of climate change debate”, New York
Times, 26.
[53] Savin, D., Alkass, S. and Fazio, P. (1996), “Construction resource leveling using neural
networks”, Canadian Journal of Civil Engineering 23, 917-925.
[54] Shi, H. W. and Li, W. Q. (2008), Application of PSO-based Neural Network in Quality
Assessment of Construction Project, Los Alamitos, Ieee Computer Soc.
[55] Shi, J. J. (1999), “A neural network based system for predicting earthmoving production”,
Construction Management & Economics 17, 463-471.
Unauthenticated
Download Date | 7/31/17 11:41 PM
13
Mojtaba Maghrebi, Claude Sammut and Travis S. Waller
[56] Silva, C. A., Faria, J. M., Abrantes, P., Sousa, J. M. C., Surico, M. and Naso, D. (2005),
“Concrete Delivery using a combination of GA and ACO. Decision and Control and
European Control Conference”, CDC-ECC '05. 44th IEEE Conference on, 12-15 Dec.
2005, 7633-7638.
[57] Sonmez, R. and Rowings, J. E. (1998), “Construction labor productivity modeling with
neural networks”, Journal of Construction Engineering and Management 124, 498-504.
[58] Srichandum, S. and Rujirayanyong, T. (2010) “Production scheduling for dispatching ready
mixed concrete trucks using bee colony optimization”, Am. J. Engg. & Applied Sci 3, 823830.
[59] Szu, H., Sheng, Y. and Chen, J. (1992), “Wavelet transform as a bank of the matched
filters”, Applied optics 31, 3267-3277.
[60] Thomas, H. R. (1991), “Labor productivity and work sampling: The bottom line”, Journal
of Construction Engineering and Management 117, 423-444.
[61] Thomas, H. R. and Daily, J. (1983), “Crew performance measurement via activity
sampling”, Journal of Construction Engineering and Management 109, 309-320.
[62] Thomas, H. R., Guevara, J. M. and Gustenhoven, C. T. (1984), “Improving productivity
estimates by work sampling”, Journal of construction engineering and management 110,
178-188.
[63] Thomas, H. R. and Sakarcan, A. S. (1994), “Forecasting labor productivity using factor
model”, Journal of Construction Engineering and Management 120, 228-239.
[64] Worrell, E., Price, L., Martin, N., Hendriks, C. and Meida, L. O. (2001), “Carbon dioxide
emissions from the global cement industry 1”, Annual Review of Energy and the
Environment 26, 303-329.
[65] Wu, D.-P., Lu, M. and Zhang, J.-P. (2005), “Efficient optimization procedures for
stochastic simulation systems. Machine Learning and Cybernetics”, Proceedings of 2005
International Conference on, 2005. IEEE, 2895-2900.
[66] Yan, S., Lai, W. and Chen, M. (2008), “Production scheduling and truck dispatching of
ready mixed concrete”, Transportation Research Part E: Logistics and Transportation
Review 44, 164-179.
[67] Yan, S., Lin, H. and Jiang, X. (2012), “A planning model with a solution algorithm for
ready mixed concrete production and truck dispatching under stochastic travel times”,
Engineering Optimization 44, 427-447.
[68] Zahraee, S. M., Hatami, M., Mohd Yusof, N., Mohd Rohani, J. and Ziaei, F. (2013),
“Combined Use of Design of Experiment and Computer Simulation for Resources Level
Determination in Concrete Pouring Process”, Jurnal Teknologi 64.
[69] Zor, I., Guzel, A., Ozyurek, H. and Erdogan, E. (2012), “Abnormal Effects of Weekdays on
Forecasting Stock Prices by Neural Networks”, European Journal of Social Sciences 29,
244-259.
14
Unauthenticated
Download Date | 7/31/17 11:41 PM